Cascade adaptive predictive controller design based on multiple models (Case study: GILAN combined cycle)

This paper presents a cascade adaptive predictive controller (APC) for an industrial application, namely, drum of GILAN combine cycle steam power plant. In order to achieve an accurate model for the drum, real data from GILAN power plant is used. The main objective of the controller design is to achieve an acceptable performance for regulating water level deviation of the drum. APC outer loop of the designed cascade controller adapts with the process dynamics variations, namely, load changes (as a measured disturbance) as well as pressure changes (as an unmeasured disturbance) effectively. The drum water level is maintained in the safe region. In order to achieve the above features, a bank of possible nonlinear models is employed. Each model is used to re-initialize the adaptive predictive controller every time. Simulation results for nonlinear drum model of GILAN combine cycle power plant show that the designed cascade adaptive predictive controller has better performance compared to the well tuned cascade proportional integral controllers.